Many Mayan languages exhibit clause type-based split ergativity in which ergative morphology (Set A) is extended to mark intransitive subjects in certain aspectless embedded clauses (“extended ergativity”). Mamean-branch Mayan languages like Mam exhibit so-called “super-extended ergativity” (England 2017), in which Set A is extended to mark all arguments in aspectless embedded clauses. In this paper, we present an analysis of the super-extended ergative pattern in Mam, which rests on three key assumptions. First, we follow a long-standing intuition in Mayan linguistics that aspectless embeddings are in fact possessed nominalizations (Comrie 1978, Laren & Norman 1979, Coon 2013, a.o.). Second, we follow much work on high-absolutive syntax in assuming that in Mam, transitive objects consistently raise above ergative subjects (Campana 1992, Coon et al. 2014, a.o.). Third, we assume that possessive structure in Mayan parallels high-absolutive syntax in also involving raising of the possessum over the possessor (Coon 2013a, Deal & Royer 2023). With these assumptions in hand, we show that object raising in Mam feeds configurations in which the transitive object is also the notional possessor of the relevant nominalization, explaining the super-extended ergative pattern in Mam. We then turn to explain the intra-family variation between Mayan languages displaying extended (e.g. Q’anjob’al) vs. super-extended ergativity (e.g. Mam). We locate this variation in the size of possible nominalizations: some languages only allow for smaller nominalizations, which essentially bleed the possibility for object raising, and thus super-extended ergativity.